On Mon, 20 Feb 2017 23:12:27 +0000
Mark Thompson <[email protected]> wrote:
> >> + switch (src_fc->sw_format) {
> >> + case AV_PIX_FMT_YUV420P: va_fourcc = VA_FOURCC_IYUV; break;
> >> + case AV_PIX_FMT_NV12: va_fourcc = VA_FOURCC_NV12; break;
> >> + case AV_PIX_FMT_P010: va_fourcc = VA_FOURCC_P010; break;
> >> + case AV_PIX_FMT_RGBA: va_fourcc = VA_FOURCC_RGBA; break;
> >> + case AV_PIX_FMT_BGRA: va_fourcc = VA_FOURCC_BGRA; break;
> >> + case AV_PIX_FMT_ARGB: va_fourcc = VA_FOURCC_ARGB; break;
> >> + case AV_PIX_FMT_ABGR: va_fourcc = VA_FOURCC_ABGR; break;
> >
> > Don't we (or should we) have a table for mapping those somewhere?
>
> Yes. I guess it could be an ff_* function in hwcontext_vaapi.c? (I've been
> trying to avoid that sort of dependency, but maybe it doesn't matter. The
> miscellaneous functions can just go at the bottom of hwcontext_internal.h.)
Doesn't sound like a problem. Better to reduce code duplication.
> > Otherwise I can't comment much.
>
> I'm not sure whether to be glad or not about that... (This is probably the
> craziest patch in the series.)
Probably not glad. I can't comment because I don't know much about
OpenCL. I mean, for the most part it seems to make sense (the way it
uses drm prime/dmabuf is similar to what you'd do in OpenGL), but the
details evade me.
_______________________________________________
libav-devel mailing list
[email protected]
https://lists.libav.org/mailman/listinfo/libav-devel
